U.S. patent application number 11/600524 was filed with the patent office on 2007-06-07 for log-likelihood ration (llr) generating apparatus and method in mimo antenna communication system.
This patent application is currently assigned to SAMSUNG ELECTRONICS CO., LTD.. Invention is credited to Keun-Chul Hwang, Young-Heon Kwon, Soon-Young Yoon.
Application Number | 20070127589 11/600524 |
Document ID | / |
Family ID | 38118713 |
Filed Date | 2007-06-07 |
United States Patent
Application |
20070127589 |
Kind Code |
A1 |
Hwang; Keun-Chul ; et
al. |
June 7, 2007 |
Log-likelihood ration (LLR) generating apparatus and method in MIMO
antenna communication system
Abstract
A Log-Likelihood Ratio (LLR) generating apparatus and method in
a communication system using a spatial multiplexing scheme. The
reception method includes acquiring at least one estimated transmit
vector by demodulating a receive vector using at least one Multiple
Input Multiple Output (MIMO) detection process; selecting one of
the at least one estimated transmit vector as an optimum estimated
transmit vector; calculating LLRs with respect to the optimum
estimated transmit vector; calculating a weight to be applied to
each of elements constructing the optimum estimated transmit
vector; and applying the weight to each of the calculated LLRs.
Accordingly, the present invention can generate the LLRs with high
reliability similar to the LLRs of a Maximum Likelihood receiver by
applying the weight to the LLRs generated from the estimated
transmit vector.
Inventors: |
Hwang; Keun-Chul;
(Seongnam-si, KR) ; Yoon; Soon-Young; (Seoul,
KR) ; Kwon; Young-Heon; (Seongnam-si, KR) |
Correspondence
Address: |
THE FARRELL LAW FIRM, P.C.
333 EARLE OVINGTON BOULEVARD
SUITE 701
UNIONDALE
NY
11553
US
|
Assignee: |
SAMSUNG ELECTRONICS CO.,
LTD.
Suwon-si
KR
|
Family ID: |
38118713 |
Appl. No.: |
11/600524 |
Filed: |
November 16, 2006 |
Current U.S.
Class: |
375/267 |
Current CPC
Class: |
H04L 25/067 20130101;
H04B 7/0413 20130101; H04B 7/0848 20130101 |
Class at
Publication: |
375/267 |
International
Class: |
H04L 1/02 20060101
H04L001/02 |
Foreign Application Data
Date |
Code |
Application Number |
Nov 16, 2005 |
KR |
2005/0109584 |
Claims
1. A receiver in a communication system using a spatial
multiplexing scheme, the receiver comprising: a Multiple Input
Multiple Output (MIMO) detector which generates at least one
estimated transmit vector by demodulating a receive vector using at
least one MIMO detection process; a reliability predictor which
calculates a weight to be applied to each of elements constructing
the at least one estimated transmit vector; and a Log-Likelihood
Ratio (LLR) generator which calculates LLRs by selecting an optimum
estimated transmit vector from the at least one estimated transmit
vector generated at the MIMO detector, multiplies the calculated
LLRs by the weight calculated at the reliability predictor, and
outputs a product.
2. The receiver of claim 1, wherein the reliability predictor
calculates the weight using the at least one estimated transmit
vector.
3. The receiver of claim 1, wherein the reliability predictor
calculates a weight to be applied to an n-th element using a number
of estimated transmit vectors having the n-th element which is the
same as an n-th element of the optimum estimated transmit
vector.
4. The receiver of claim 1, wherein the reliability predictor
calculates the weight using a channel coefficient matrix.
5. The receiver of claim 1, wherein the LLR generator comprises: a
vector selector which calculates distances between the at least one
estimated transmit vector generated at the MIMO detector and the
receive vector, respectively, and selects and outputs an estimated
transmit vector having a minimum distance as the optimum estimated
transmit vector; a calculator which calculates an LLR with respect
to each of elements constructing the optimum estimated transmit
vector selected at the vector selector; and a multiplier which
multiplies each of the LLRs calculated at the calculator by the
weight calculated at the reliability predictor.
6. The receiver of claim 5, wherein the calculator comprises: a
variable generator which generates a soft decision variable with
respect to each of the elements constructing the optimum estimated
transmit vector selected at the vector selector; and an LLR
calculator which calculates an LLR with respect to each of encoded
symbols by executing an LLR calculation process with the soft
decision variables generated at the variable generator.
7. The receiver of claim 6, wherein the soft decision variable is
calculated based on the following equation: d n = h n H ( y - i = 1
, i .noteq. n N T .times. h i .times. x ~ i ( opt ) ) ##EQU12##
where d.sub.n denotes a soft decision variable corresponding to a
transmit symbol of an n-th transmit antenna, h.sub.n denotes an
n-th column vector of a channel coefficient matrix H, and {tilde
over (x)}.sub.i.sup.(opt) denotes an i-th element of the optimum
estimated transmit vector {tilde over (x)}.sub.opt.
8. The receiver of claim 6, wherein the LLR calculator calculates
the LLRs using an LLR calculation process of a single input and
single output (SISO) system.
9. The receiver of claim 1, further comprising: a deinterleaver
which deinterleaves and outputs LLRs to which the weight calculated
at the multiplier is applied; and a channel decoder which restores
information bit streams by soft-decision decoding the LLRs output
from the deinterleaver.
10. A reception method in a communication system using a spatial
multiplexing scheme, the method comprising: acquiring at least one
estimated transmit vector by demodulating a receive vector using at
least one Multiple Input Multiple Output (MIMO) detection process;
selecting one of the estimated transmit vectors as an optimum
estimated transmit vector; calculating Log-Likelihood Ratios (LLRs)
with respect to the optimum estimated transmit vector; calculating
a weight to be applied to each of elements constructing the optimum
estimated transmit vector; and applying the weight to each of the
calculated LLRs.
11. The reception method of claim 10, wherein the weight is
calculated using the at least one estimated transmit vector.
12. The reception method of claim 10, wherein a weight to be
applied to an n-th element is calculated using a number of
estimated transmit vectors having the n-th element which is the
same as an n-th element of the optimum estimated transmit
vector.
13. The reception method of claim 10, wherein the weight is
calculated using a channel coefficient matrix.
14. The reception method of claim 10, wherein the selecting step
comprises: calculating distances between the at least one estimated
transmit vector and the receive vector, respectively; and selecting
an estimated transmit vector having a minimum distance among the
calculated distances as the optimum estimated transmit vector;
15. The reception method of claim 10, wherein the calculating LLR
step comprises: generating a soft decision variable with respect to
each of the elements constructing the optimum estimated transmit
vector; and generating an LLR with respect to each of encoded
symbols by applying an LLR calculation process to each of the soft
decision variables.
16. The reception method of claim 15, wherein the soft decision
variable is calculated based on the following equation: d n = h n H
( y - i = 1 , i .noteq. n N T .times. h i .times. x ~ i ( opt ) )
##EQU13## where d.sub.n denotes a soft decision variable
corresponding to a transmit symbol of an n-th transmit antenna,
h.sub.n denotes an n-th column vector of a channel coefficient
matrix H, and {tilde over (x)}.sub.i.sup.(opt) denotes an i-th
element of the optimum estimated transmit vector {tilde over
(x)}.sub.opt.
17. The reception method of claim 10, further comprising:
deinterleaving and outputting LLRs to which the weight is applied;
and restoring information bit streams by soft-decision decoding the
deinterleaved LLRs.
Description
PRIORITY
[0001] This application claims priority under 35 U.S.C. .sctn. 119
to an application filed in the Korean Intellectual Property Office
on Nov. 16, 2005 and assigned Serial No. 2005-109584, the contents
of which are incorporated herein by reference.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The present invention generally relates to a multi-antenna
communication system, and in particular, to an apparatus and method
of generating a Log-Likelihood Ratio (LLR) of high reliability in a
communication system using a spatial multiplexing scheme.
[0004] 2. Description of the Related Art
[0005] Recently, with rapid growth of the wireless mobile
communication market, various multimedia services in the wireless
environment are becoming more heavily demanded. In particular, mass
transmission data and rapid data delivery are progressing. Thus, an
urgent task is to find a method of efficiently using limited
frequencies. To respond to this, a new transmission technique using
a multi-antenna is desired. By way of example of the new
transmission technique, a Multiple Input Multiple Output (MIMO)
system using a multi-antenna is being used.
[0006] The MIMO technique, which uses a multi-antenna at the
transmitter and the receiver respectively, can increase the channel
transmission capacity in proportion to the number of the antennas
without additional frequencies or transmit power allocation,
comparing to a system using a single antenna. Thus, in recent
years, active research is being conducted on the MIMO
technique.
[0007] The Multi-antenna techniques are divided largely to a
spatial diversity scheme which improves the transmission
reliability by acquiring a diversity gain corresponding to the
product of the numbers of transmit and receive antennas, a Spatial
Multiplexing (SM) scheme which increases the data rate by
transmitting a plurality of signal streams at the same time, and a
combination scheme of the spatial diversity and the SM.
[0008] When transmitters send different data streams using an SM
scheme of the multi-antenna techniques, interference occurs between
the data transmitted simultaneously. Hence, the receiver detects a
signal using a Maximum Likelihood (ML) receiver by taking account
of the influence of the interference signal, or detects the signal
after rejecting the interference. The interference cancellation
schemes include a Zero Forcing scheme, a Minimum Mean Square Error
(MMSE) scheme, and so forth. In a general SM scheme, there is the
trade-off between the receiver performance and the computational
complexity of the receiver. Thus, active research is conducted on a
reception process which can achieve performance approximate to an
ML receiver with a low computational complexity of the
receiver.
[0009] In the mean time, it is known that decoding by providing a
soft decision value to a channel decoder is beneficial in terms of
the performance, rather than providing a hard decision value of
encoded bits. A input soft decision value of the decoder, which is
an estimated value of modulated symbols transmitted in the channel,
uses a Log-Likelihood Ratio (LLR) value. Accordingly, a receiver of
an SM scheme uses a process which generates an optimum LLR from a
corresponding reception process, besides a low-complexity reception
process.
[0010] Conventional signal detection methods using the SM scheme
include the ML, Successive Interference Cancellation (SIC), and
Vertical Bell Labs Layered Space-Time (V-BLAST), and so forth.
[0011] To briefly explain the processes, a system model is defined.
The system of the interest is a model including N.sub.T-ary
transmit antennas and N.sub.R-ary receive antennas as shown in FIG.
1. When expressing a signal to transmit on each transmit antenna as
x.sub.m, a receive signal y at the receiver can be expressed by
Equation (1) below. It is assumed that the signal x.sub.m to
transmit on the transmit antenna is an M-QAM signal. The number of
the encoded bits that can be transmitted at a time is
N.sub.T.times.log.sub.2 (M). y=Hx+n (1)
[0012] In Equation (1), y is a receive signal vector, x is the
transmit symbol vector, and H is a channel coefficient matrix
generated between the transmit antenna and the receive antenna,
which is defined by Equation (2) below. n denotes an ambient
Gaussian noise vector. x = [ .times. x 1 , x 2 , x 3 , .times. , x
N T .times. ] T y = [ .times. y 1 , y 2 , y 3 , .times. , y N R
.times. ] T H = [ h 11 h 12 h 1 .times. N T h 21 h 22 h 2 .times. N
Y h N R .times. 1 h N R .times. 2 h N R .times. N T ] ( 2 )
##EQU1##
[0013] In Equation (2), the channel coefficient matrix H is
N.sub.R.times.N.sub.T matrix. The element h.sub.ij corresponding to
the i-th line and the j-th column denotes the channel response
between the j-th transmit antenna and the i-th receive antenna.
[0014] Signal detection methods using the SM scheme are arranged as
follows.
[0015] The ML scheme selects a symbol vector having the shortest
direct distance by computing the Euclidean distance, as defined
below in Equation (3), with respect to all symbol vectors in the
constellation. In other words, the ML scheme, which measures the
distance between y and Hx, determines a symbol vector having the
shortest distance as a symbol vector with the highest similarity,
that is, with the minimum error. However, it is hard to practically
realize the ML scheme because the complexity increases by raising
the length of the codeword to the power of the number of the
transmit antennas as shown in M.sup.N.sup.T(M.times.ary,|c|=M). x ^
= arg x .times. min .times. y - Hx F 2 ( 3 ) ##EQU2##
[0016] The SIC scheme cancels the interference of the received
signal by reconstructing the values of the hard decision at the
previous step. However, if the hard decision values of the previous
step suffer error, the SIC scheme aggravates the error in the next
step. Thus, in every step, the reliability of the hard decision
values deteriorates.
[0017] Accordingly, the SIC scheme needs to take account of error
propagation which is the factor of the performance degradation.
Specifically, since the decoding is performed in the order of the
transmit antenna index regardless of the channel status during
interference cancellation, interference cancellation is carried out
without removing the transmit antenna of the great signal
intensity. As a result, the performance of the transmit antenna
signal with the weak signal intensity is not considerably enhanced.
A V-BLAST process addresses this problem and shows the better
performance than an existing SIC scheme by canceling interference
in an order of transmit antennas having the greater signal
intensity.
[0018] The Modified ML (MML) scheme, by ML-decoding the symbol
vectors transmittable on the other transmit antennas, excluding a
signal transmitted with an arbitrary transmit antenna, can
determine the one signal through the simple slicing operation Q( )
as shown below in Equation (4). The MML scheme exhibits the
performance similar to the ML scheme and its computational
complexity increases by raising to the power of the number of the
transmit antennas minus 1. That is, the ML scheme computes the
Euclidean distance with respect to M.sup.N.sup.T-ary transmit
vectors, whereas the MML computes the Euclidean distance with
respect to M.sup.N.sup.T.sup.-1-ary transmit vectors and detects
the signal of the rest symbol through the slicing operation as in
Equation (4). x i = Q ( h i H h 1 2 .times. ( y - j .noteq. i
.times. h j .times. x j ) ) ( 4 ) ##EQU3##
[0019] Finally, the Recursive MML (RMML) scheme is suggested to far
more mitigate the complexity of the MML. The RMML generates a
plurality of subsystems by nulling the channel using Givens
rotation and decides the MML in the minimum unit 2.times.2 channel.
As such, the RMML scheme can mitigate the computational complexity
with the performance similar to the ML by generating the subsystems
(e.g., 3.times.3 and 2.times.2). Yet, the generation of the
multiple subsystems implies the multiple candidate transmit
vectors, which limit the complexity mitigation. In addition, since
the decision is made in the 2.times.2 subsystem right away, the
performance degradation arises like the SIC family.
[0020] Meanwhile, the LLR computation at the MIMO receiver differs
depending on the reception processes. In the MIMO environment with
the inter-signal interference, the reliability of the LLR value
quite differs depending on the MIMO reception processes. The
reliability of the LLR value directly affects the decoding
performance of the decoder. In the ML receiver which is known as
the optimum receiver amongst the various MIMO reception processes,
the optimum LLR computation is expressed by Equation (5) below. LLR
optmum .function. ( b i ) = .times. log .times. .times. P
.function. ( b i = + 1 | y ) P .function. ( b i = - 1 | y ) =
.times. log .times. .times. P .times. ( y | b i = + 1 ) .times. P
.function. ( b i = + 1 ) P .function. ( y | b i = - 1 ) .times. P
.function. ( b i = - 1 ) = .times. log .times. .times. P .function.
( y | b i = + 1 ) P .function. ( y | b i = - 1 ) = .times. log
.times. .times. x + .di-elect cons. C i + .times. P .function. ( y
| x = x + ) .times. P .function. ( x = x + ) x - .di-elect cons. C
i - .times. P .function. ( y | x = x - ) .times. P .function. ( x =
x - ) = .times. log .times. .times. x + .di-elect cons. C i +
.times. exp .function. ( - y - Hx + 2 2 .times. .sigma. 2 ) .times.
b j .di-elect cons. x + .times. P .function. ( b j ) x - .di-elect
cons. C i - .times. exp .function. ( - y - Hx - 2 2 .times. .sigma.
2 ) .times. b j .di-elect cons. x + .times. P .function. ( b j )
.apprxeq. .times. log .times. .times. max x + .di-elect cons. C i +
.times. exp .function. ( - y - Hx + 2 2 .times. .sigma. 2 ) max x -
.di-elect cons. C i - .times. exp .function. ( - y - Hx - 2 2
.times. .sigma. 2 ) = .times. 1 2 .times. .sigma. 2 .times. { min x
+ .di-elect cons. C i + .times. y - Hx + 2 - min x - .di-elect
cons. C i - .times. y - Hx - 2 } ( 5 ) ##EQU4##
[0021] In Equation (5), b.sub.i denotes an i-th bit.
P(b.sub.i=+1|y) denotes a probability of the i-th bit being `+1`
when the receive signal vector y is received, and P(b.sub.i=+1)
denotes a probability of the i-th bit being `+1`. C.sub.i.sup.+
denotes the set of x's when the i-th bit of the transmit signal
vector x is `+1`, and C.sub.i.sup.- denotes the set of x's when the
i-th bit of the transmit signal vector x is `-1`. As one can see
from Equation (5), since the LLR computation at the ML receiver has
to calculate the Euclidean distance with respect to every
combination of the transmit signal vector x, it is difficult to
adopt it for the greater number of antennas or the high-level
modulation scheme.
[0022] As discussed above, when using the SM scheme, what is
demanded is a receiver structure which has low complexity and high
reliability similar to the LLR of the ML.
SUMMARY OF THE INVENTION
[0023] An aspect of the present invention is to substantially solve
at least the above problems and/or disadvantages and to provide at
least the advantages below. Accordingly, an aspect of the present
invention is to provide a receiving apparatus and method of
generating a Log-Likelihood Ratio (LLR) with high reliability in a
system using a Spatial Multiplexing (SM) scheme.
[0024] Another aspect of the present invention is to provide a
receiving apparatus and method which has low complexity and shows
the performance similar to a Maximum Likelihood (ML) scheme in a
system using the SM scheme.
[0025] The above aspects are achieved by providing a receiver in a
communication system using a SM scheme, the receiver including a
multiple input multiple output (MIMO) detector which generates at
least one estimated transmit vector by demodulating a receive
vector using at least one MIMO detection process; a reliability
predictor which calculates a weight to be applied to each of
elements constructing the at least one estimated transmit vector;
and a LLR generator which calculates LLRs by selecting an optimum
estimated transmit vector from the at least one estimated transmit
vector generated at the MIMO detector, multiplies the calculated
LLRs by the weight calculated at the reliability predictor, and
outputs the product.
[0026] According to one aspect of the present invention, a
reception method in a communication system using a spatial
multiplexing scheme includes acquiring at least one estimated
transmit vector by demodulating a receive vector using at least one
MIMO detection process; selecting one of the estimated transmit
vectors as an optimum estimated transmit vector; calculating LLRs
with respect to the optimum estimated transmit vector; calculating
a weight to be applied to each of elements constructing the optimum
estimated transmit vector; and applying the weight to each of the
calculated LLRs.
BRIEF DESCRIPTION OF THE DRAWINGS
[0027] The above and other objects, features and advantages of the
present invention will become more apparent from the following
detailed description when taken in conjunction with the
accompanying drawings in which:
[0028] FIG. 1 illustrates a configuration of a transmitter and a
receiver in a Multiple Input Multiple Output (MIMO) antenna system
according to the present invention;
[0029] FIG. 2 illustrates a detail configuration of the MIMO
demodulator of FIG. 1;
[0030] FIG. 3 illustrates a detail configuration of the LLR
generator of FIG. 2;
[0031] FIG. 4 illustrates a procedure of generating LLRs fed to the
channel decoder in the MIMO antenna system according to the present
invention;
[0032] FIG. 5 illustrates a 16QAM constellation according to the
present invention; and
[0033] FIG. 6 is a graph comparing a packet error rate of various
MIMO reception processes.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0034] Preferred embodiments of the present invention will be
described herein below with reference to the accompanying drawings.
In the following description, well-known functions or constructions
are not described in detail since they would obscure the invention
in unnecessary detail.
[0035] Hereinbelow, descriptions explain a receiver and method of
generating a Log-Likelihood Ratio (LLR) with high reliability in a
system using a Spatial Multiplexing (SM) scheme according to the
present invention.
[0036] FIG. 1 shows a transmitter and a receiver in a Multiple
Input Multiple Output (MIMO) antenna according to the present
invention. The transmitter includes a channel encoder 100, an
interleaver 110, a modulator 120, a demultiplexer (DEMUX) 130, and
a plurality of transmit antennas 140-1 through 140-N.sub.T. The
receiver includes a plurality of receive antennas 150-1 through
150-N.sub.R, a MIMO demodulator 160, a deinterleaver 170, and a
channel decoder 180.
[0037] In the transmitter, the channel encoder 100 outputs encoded
symbols by coding information bit streams to be transmitted at a
given coding rate. When the number of the input information bits is
k and the coding rate is R, the number of the output symbols is
k/R. By way of example, the channel encoder 100 can employ a
convolutional encoder, a turbo encoder, a Low Density Parity Check
(LDPC) encoder, and the like.
[0038] The interleaver 110 interleaves and outputs the symbols from
the channel encoder 100 according to an interleaving rule to make
them robust to burst error.
[0039] The modulator 120 modulates and outputs the symbols
interleaved at the interleaver 110 using a certain modulation
scheme. In other words, the modulator 120 outputs a complex signal
by mapping signal points into a constellation according to a
certain mapping scheme. For instance, the modulation schemes
include Binary Phase Shift Keying (BPSK) which maps one bits (s=1)
to one complex signal, Quadrature Phase Shift Keying (QPSK) which
maps two bits (s=2) to one complex signal, 8-ary Quadrature
Amplitude Modulation (8QAM) which maps three bits (s=3) to one
complex signal, and 16QAM which maps four bits (s=4) to one complex
signal.
[0040] The DEMUX 130 demultiplexes the complex signals fed from the
modulator 120 and transmits them on the N.sub.T-ary transmit
antennas. Although not show in the drawings, for instance, when
using an Orthogonal Frequency Division Multiplexing (OFDM)
communication scheme, the plurality of the streams output from the
DEMUX 130 are OFDM-modulated and the OFDM-modulated signal is
transmitted over the corresponding antenna by air after passing
through Radio Frequency (RF) processing to enable transmission by
air. It is assumed that the transmit vector transmitted on the
plurality of the transmit antennas 140-1 through 140-N.sub.T is
x=[x.sub.1, x.sub.2, . . . , x.sub.N.sub.T].
[0041] Next, in the receiver, the plurality of the receive antennas
150-1 through 150-N.sub.R receives the signals transmitted from the
plurality of the transmit antennas 140-1 through 140-N.sub.T.
Although not shown in the drawing, the RF signals received through
the receive antennas 150-1 through 150-N.sub.R are converted to
baseband sample data, respectively. The sample data is
OFDM-demodulated and then fed to the MIMO demodulator 160. It is
assumed that the receive vector input to the MIMO demodulator 160
is y=[y.sub.1, y.sub.2, . . . , y.sub.N.sub.R].
product. Operation of the MIMO demodulator 160 will be described by
referring to FIG. 2.
[0042] The deinterleaver 170 deinterleaves and outputs the LLRs fed
from the MIMO demodulator 160 according to a given rule. The
channel decoder 180 restores the information bit streams by
soft-decision decoding the LLRs from the deinterleaver 170.
[0043] As shown in FIG. 2, the MIMO demodulator 160 includes a
plurality (K-ary) of MIMO detectors 201-1 through 201-K, a
reliability predictor 203, and an LLR generator 205.
[0044] The plurality of the MIMO detectors 201-1 through 201-K
generates one or more estimated vectors (transmit vectors) by
demodulating a receive vector y using different MIMO detection
schemes. For instance, the MIMO detectors can be configured by
differently applying a detection order in the MMSE-OSIC (Ordered
SIC) process. In addition, an ascending part and a descending part
detected in the Sorted MML can construct two different MIMO
detectors. Likewise, it is assumed that the MIMO detectors 201-1
through 201-K are constructed using a variety of present or future
MIMO detection processes. It is assumed that the estimated vectors
output from the MIMO detectors 201-1 through 201-K are {circumflex
over (x)}.sub.1, {circumflex over (x)}.sub.2, . . . , {circumflex
over (x)}.sub.K.
[0045] The reliability predictor 203 predicts a reliability with
respect to each of elements (estimated symbols) constructing an
optimum estimated vector, and calculates and outputs a weight to
multiply by each of the elements (symbols). In doing so, a high
weight is applied to a high-reliability element (symbol) and a low
weight is applied to a low-reliability symbol. Equation (6)
expresses a simple example of the weight generated at the
reliability predictor 203. and outputs a weight to multiply by each
of the elements (symbols). In doing so, a high weight is applied to
a high-reliability element (symbol) and a low weight is applied to
a low-reliability symbol. Equation (6) expresses a simple example
of the weight generated at the reliability predictor 203. w n = { 1
, when .times. .times. x ~ n ( opt ) = x ~ n ( 1 ) = = x ~ n ( K )
0 , otherwise ( 6 ) ##EQU5##
[0046] In Equation (6), {tilde over (x)}.sub.n.sup.(k) denotes an
n-th element of an k-th estimated vector {tilde over (x)}.sub.k.
That is, when the elements at the same position in every estimated
vector have the equal value, the weight for the elements is set to
`1`. Otherwise, the weight is set to `0`.
[0047] Equation (7) expresses another example of the weight
generated at the reliability predictor 203. w n = number .times.
.times. of .times. .times. the .times. .times. same .times. .times.
symbol with .times. .times. x ~ n ( opt ) .times. .times. among
.times. { x ~ n ( k ) | k - 1 , 2 , .times. , K } K ( 7 )
##EQU6##
[0048] In Equation (7), the weight can be interpreted as the
probability that the n-t h element of the K-ary estimated vectors
is the same as the n-th element of {tilde over (x)}.sub.opt (the
optimum vector selected from the estimated vectors). This can be
interpreted as the reliability when the n-th element (or the
estimated symbol) {tilde over (x)}.sub.n.sup.(opt) is the symbol
that is correctly determined.
[0049] Meanwhile, the LLR generator 205 generates LLRs by selecting
the optimum estimated vector from the K-ary estimated vectors of
the MIMO detectors 201-1 through 201-K, multiplies the generated
LLRs by the weight acquired at the reliability predictor 203, and
outputs the product. Operation of the LLR generator 205 will be
explained by referring to FIG. 3.
[0050] The LLR generator 205 according to the present invention
includes a minimum distance vector selector 301, a soft decision
variable generator 303, a plurality of LLR calculators 305-1
through 305-K, a plurality of multipliers 307-1 through 307-K, and
a multiplexer (MUX) 309.
[0051] The minimum distance vector selector 301 calculates the
Euclidean distance between the K-ary estimated vectors ({circumflex
over (x)}.sub.1, {circumflex over (x)}.sub.2 . . . , {circumflex
over (x)}.sub.K) from the MIMO detectors 201-1 through 201-K and
the receive vector y, respectively, selects and outputs an optimum
estimated vector {tilde over (x)}.sub.opt corresponding to the
minimum Euclidean distance. The modeling of the operation of the
minimum distance vector selector 301 is expressed as in Equation
(8) below. x ~ opt = x ~ arg .times. .times. min k .times. y - H
.times. x k ~ , k = 1 , 2 , .times. , K ( 8 ) ##EQU7##
[0052] The soft decision variable generator 303 computes and
outputs a soft decision variable for each estimated symbol with
respect to the optimum estimated vector {tilde over (x)}.sub.opt
fed from the minimum distance vector selector 301 based on the
following Equation (9). d n = h n H ( y - i = 1 , i .noteq. n N T
.times. h i .times. x ~ i ( opt ) ) ( 9 ) ##EQU8##
[0053] In Equation (9), d.sub.n denotes the soft decision variable
corresponding to the transmit symbol of the n-th transmit antenna,
h.sub.n denotes the n-th column vector of the channel coefficient
matrix H, and {tilde over (x)}.sub.i.sup.(opt) denotes the i-th
element of {tilde over (x)}.sub.opt.
[0054] Since the calculated soft decision variable is a scalar
value, rather than a vector, the LLR can be acquired using the LLR
calculation process of the well-known Single Input Single Output
(SISO) system.
[0055] Accordingly, the LLR calculators 305-1 through 305-K
calculate and output the LLR with respect to each of the encoded
symbols (bits) mapped to one signal point using the input soft
decision variable d.sub.n, respectively. In case of the 16QAM as
shown in FIG. 5, the LLR calculation process of the SISO system can
be defined as in Equation (10) below. LLR .function. ( b n , 1 ) =
4 .times. a .times. h n 2 2 .times. .sigma. 2 .times. { real
.times. .times. ( d n ' ) , when .times. real .times. .times. ( d n
' ) .ltoreq. 2 .times. a 2 .times. ( real .times. .times. ( d n ' )
- a ) , when .times. real .times. .times. ( d n ' ) > 2 .times.
a 2 .times. ( real .times. .times. ( d n ' ) + a ) , when .times.
real .times. .times. ( d n ' ) < - 2 .times. a LLR .function. (
b n , 2 ) = 4 .times. a .times. h n 2 2 .times. .sigma. 2 .times. (
2 .times. a - real .times. .times. ( d n ' ) ) LLR .function. ( b n
, 3 ) = 4 .times. a .times. h n 2 2 .times. .sigma. 2 .times. {
image .times. .times. ( d n ' ) , when .times. image .times.
.times. ( d n ' ) .ltoreq. 2 .times. a 2 .times. ( image .times.
.times. ( d n ' ) - a ) , when .times. image .times. .times. ( d n
' ) > 2 .times. a 2 .times. ( image .times. .times. ( d n ' ) +
a ) , when .times. image .times. .times. ( d n ' ) < - 2 .times.
a LLR .function. ( b n , 4 ) = 4 .times. a .times. h n 2 2 .times.
.sigma. 2 .times. ( 2 .times. a - image .times. .times. ( d n ' ) )
( 10 ) ##EQU9##
[0056] In Equation (10), b.sub.n,j denotes the i-th encoded symbol
of the n-th transmit antenna, real(d'.sub.n) denotes the real
component, and image(d'.sub.n) denotes the imaginary component.
d'.sub.n is expressed as in Equation (11) below. d n ' = d n h n 2
( 11 ) ##EQU10##
[0057] In other words, the LLR calculators 305-1 through
305-N.sub.T calculate and output log.sub.2(M)-ary LLRs
corresponding to the n-th transmit antenna, respectively.
[0058] The plurality of the multipliers 307-1 through 307-N.sub.T
multiplies the LLR from the corresponding LLR generator by the
weight from the reliability predictor 203, and outputs its
product.
[0059] The MUX 309 converts the LLRs weighted at the multipliers
307-1 through 307-N.sub.T to serial data and outputs the serial
data to the deinterleaver 170.
[0060] FIG. 4 shows a procedure of generating LLRs fed to the
channel decoder in the MIMO antenna system according to the present
invention. The receiver checks whether a signal is received in step
401. When the signal is received, the receiver acquires a plurality
of estimated vectors (transmit vectors) by demodulating the receive
vector y using a plurality of MIMO detection processes in step 403.
Note that the use of the different MIMO detection processes does
not always imply that the estimated vectors {circumflex over
(x)}.sub.1, {circumflex over (x)}.sub.2, . . . , {circumflex over
(x)}.sub.K are different from one another.
[0061] After acquiring the plurality of the estimated vectors, the
receiver calculates the Euclidean distances between the receive
vector y and the estimated vectors and selects an estimated vector
having the minimum Euclidean distance in step 405.
[0062] Next, the receiver generates a soft decision variable for
each element constructing the selected vector {tilde over
(x)}.sub.opt in step 407. The soft decision variable corresponding
to the transmit symbol of the n-th transmit antenna can be
calculated based on Equation (9).
[0063] As such, after generating the soft decision variable for
each of the elements constructing the estimated vector, the
receiver generates log.sub.2(M)-ary LLRs corresponding to the
relevant transmit antennas using the soft decision values in step
409. In doing so, the receiver generates the LLRs using the LLR
calculation process of the SISO system.
[0064] In the mean time, the receiver predicts the reliability of
the generated LLRs and calculates a weight to be applied to each
LLR according to its predicted reliability in step 411. The weight
can be acquired using the plurality of the estimated vectors
generated using the MIMO detection processes as explained in
connection with Equations (6) and (7).
[0065] After calculating the weight to be applied to each LLR, the
receiver multiplies the LLRs to be fed to the channel decoder by
the weights, respectively in step 413 and then terminates the
process of the present invention.
[0066] The present invention generates a plurality of the estimated
vectors using a plurality of the MIMO detection processes and
applies an LLR generation method to one of the estimated vectors.
However, according to the present invention, one estimated vector
can be generated using one MIMO detection process and an LLR
generation method can be applied to the estimated vector. Since the
suggested reliability prediction (weight calculation) method of the
present invention is not applicable, the weight for the LLR is
calculated using the channel coefficient matrix (H) by way of
example.
[0067] FIG. 6 is a graph comparing a packet error rate of various
MIMO reception processes. A link-level simulation is conducted and
the performance is compared using the Packet Error Rate (PER). It
is assumed that the number of the transmit/receive antennas is
four, respectively, the OFDM system has 64 sub-carriers, and that
the transmit channel environment is a 9 tap frequency selective
channel without correlation. In addition, a packet consists of
10240 bits (10 OFDM symbols) and the 16QAM modulation scheme is
adopted.
[0068] As shown in FIG. 6, the performance is compared among
MMSE-OSIC, Sorted MML, and QRM-MLD (QR decomposition and the
M-algorithm--Maximum Likelihood Detection) of NTT-DoCoMo (a
Jananese telecommunication company).
[0069] The suggested LLR generation method of the present invention
is applied only to the Sorted MML and the weight is defined as in
Equation (12) below. w n = { 1 , when .times. .times. x AS = x DS 1
/ 8 , otherwise ( 12 ) ##EQU11##
[0070] In Equation (12), x.sub.AS and x.sub.DS denote a decision
value of the branch detected in an ascending order and an
descending order of the Sorted MML, respectively, and the value 1/8
is an optimum value acquired from the experiment.
[0071] As one can see from the graph, at 1% Packet Error Rate
(PER), the present invention (Sorted MML (soft decision)) has about
a 4.0 dB gain, comparing to the related art (Sorted MML (hard
decision)) using the hard decision. In comparison to the soft
decision of the QRM-MLD scheme, the present invention exhibits the
performance enhanced by about 2.0 dB. In comparison to the soft
decision of MMSE-OSIC which is generally known as the simple MIMO
reception process, the present invention exhibits a performance
enhancement by about 7.5 dB.
[0072] In light of the forgoing as set forth above, the present
invention can generate the LLRs with the high reliability similar
to the LLRs of the ML receiver by applying a weight to the LLRs
generated from a transmit vector estimated in a system using an SM
scheme.
[0073] While the invention has been shown and described with
reference to certain preferred embodiments thereof, it will be
understood by those skilled in the art that various changes in form
and details may be made therein without departing from the spirit
and scope of the invention as defined by the appended claims.
* * * * *